Response characteristics of karst groundwater to rainstorm and the multiple regression prediction model in Dachengqiao, Ningxiang county, Hunan Province

The groundwater activity in the karst mining area can cause serious geological hazards. It is the key to solve the problem of frequent disastersto understand the response characteristics of groundwater to extreme climate conditions. The Dachengqiao area was selected as the study area, which is about 2 km away from the Meitanba coal mine.The stratigraphic lithology in Dachengqiao is mainly the calcareous conglomerate of Cretaceous Baihuating formation, followed by limestone of Permian Qixia formation and Maokou formation.Affecting by the mining drainage karst aquifer,the maximum water level of Dachengqiao is more than 90 m deep.The main runoff direction of groundwater flows from south to north, and is discharged to the mine. In this paper,through the on-site monitoring methods such as the high frequency monitoring of karst groundwater level and rainfall,the dynamics of karst groundwater level and itsresponse process to heavy rainfall are studied,and the characteristics of ground water internal relationship are discussed and a regression predition model is established. The results indicated that karst groundwater level responded quicker to rainfall in rainy season, and the amplitude of groundwater level is larger compared with that in the dry season. Under the rainstorm condition, the dynamic response of karst groundwater level to the rapid pipeline flow is a steep rise-steepfall,with the variation range of 13.6 to 42.8 m.Whereas the dynamic characteristics of groundwater level supplied by fracture and pore water are slow rise-slow fall,withan annual variation of 1.1 m. The lag time of groundwater level response to rainstrom in therainy season ranged from 1.2 to 4 h, which is 1/2 to 2/3 shorter than that under the same conditions,and groundwater level variation increased by 1.7 to 4.7 times. In dry season,the lagtime of groundwater level response to rainstorm ranged from 4.2 to 13.2 h, which is extended by 3 to 8 h, and groundwater level variation is reduced by 5 to 11 times. According to the research, groundwater level dynamic characteristic was controlled by thickness and water saturation of vadose zone, which indicated thickness of vadose zone presented a negative relationship with response lag of groundwater level, and water saturation of aeration zone presented a positive relationship with groundwater level variation. The optimum regression prediction model is established through correlation analysis, fitting optimization, variance and residual analysis, and regression coefficientt test, which indicates as Hi=14.149+0.308y1+0.379y2-0.023Pi+0.051Ti. It is proved that the model can effectively predict the water level change process of monitoring sites under the condition of rainstorm.